Alex Rampell - Investing in Operating Systems


2x2 Matrix

🏆 Right🤷‍♂️ Wrong
📣 Consensus✅ good for venture capital❌ always bad
🕵️ Non-consensus✅ good for public equities ❌ always bad

[In VC], lean into positive selection and avoid adverse selection.

In public equities, you want to be right against the consensus. In private equities, you want to be right with the consensus (because you need other folks to lead rounds after you).

Call-options and Bonds

There are no equities. Every investment is either an out of the money call-option or a bond.

there’s no such thing as an equity. There’s only a bond and a call option. And a lot of what private markets invest in, in particular, are out of the money call options. All these call options are not in the money or at the money.

Invest in “operating-system”-like mechanics

How can I identify operating system-like products?:

  • these products have high retention rates (nearly 100%)
  • they are often software products that keep track of everything
  • they are products with very high usage (daily active usage usually)
  • these products essentially “run the business

What’s the goal and advantage of operating system-like products?:

Essentially, you want to control distribution because it gives you cross-selling abilities. This leads to a permanent upscale ability, i.e. you can continuously expand your feature set and add more value-add products without depending on distribution by other companies.

You can classify operating system-like products as:

  • horizontal (e.g. Quickbooks, which offers accounting software across all business verticals), and
  • vertical (e.g. Toast, which for offers point of sale software for restaurants).

You can think of most businesses as:

  1. Transaction-based (e.g, Peloton, which sells exercise equipment),
  2. Ad-based (e.g. Facebook), or
  3. Embedded financial services (e.g. Uber payroll, checking accounts, debit cards, …), which are not officially fintech businesses but could cross-sell certain supplementary financial services, because they controls the distribution, or Toast, which is a restaurant software business, but actually makes money with adjacent fintech services

Controlling distribution is difficult, so if you are a startup, you want to find a trojan horse product that gets you fast adoption (or a distribution wedge as Alex calls it) and cross-sell other products later.

Revolutionising lending

In a nutshell, you can think of credit risk as a function of users willingness and ability to pay.

If you wanted to revolutionise lending you would want to tightly link lending to payroll, i.e. imagine if users could choose that part of their payroll would go directly towards repaying loans and eliminate their unwillingness to pay, smart contracts are essentially that.

One of the biggest opportunities is to more tightly integrate lending with borrowers’ cash flows, such as payroll (for consumers) and revenues (for businesses). In a nutshell, you want to solve working capital issues with better data.

Rates are generally too high in the world, it’s too much “one size fits all” when we actually have good enough data to provide better interest rates. You’d think reducing interest rates = less profit, but better rates might actually expand the pie (i.e. extend credit to more folks) and turn into a volume game.

Customers generally prefer to have bundled services with as little friction and decision making as possible. If you bundle financial services like lending more tightly, you can underwrite better. Because you are tight (close to the source and thus more certain of payment), you have more data and can price better for customers. This can be a very compelling proposition. In short, if you are an embedded finacial services firm, you can cross-sell financial services on the basis of better access to data and because you control distribution.

Some of the features might be:

  • choosing your payday
  • link payroll to loan repayments
  • lower rates because of better data (certainty around payment)

Buy Now Pay Later could disrupt payment networks

BNPL offers clear benefits for consumers and clear benefits for merchants (this can be a very dangerous dynamic for incumbent payment networks).

BNPL is has the potential to be parallel network with consumers and merchants, and item specific information (SKU-level data). This is particularly compelling because merchants are not fans of the fees charged by incumbents like Visa or Mastercard.

The emphasis on the parallel network is important, because, for example, PayPal is not a different network, it’s just an abstraction layer on top of the incumbent payment network and just ends up charging my credit card.

In finance, distribution is very powerful, so competitive dyanimcs usually boil down to distribution speed (incumbents) vs innovation speed (startups).

According to Alex, Marqeta and Plaid are infrastructure layers.

Show Notes

Show notes from here:

[00:03:32] - [First question] - Lean into positive selection and avoid adverse selection

[00:07:48] - Thoughts on growing capital formation in private markets

[00:14:01] - Why it’s useful for investors to think in terms of bonds and call options instead of equity

[00:18:39] - Doing more with less and hunting for operating systems to invest in

[00:28:08] - His views on infrastructure and the presentation layer conundrum

[00:33:32] - The sequencing involved in building an operating system over time

[00:40:11] - Rise of the creator class and the coming tailwind post-cloud technology; the rise of the solopreneur 

[00:43:32] - The pig joke and his thoughts on the FinTech space

[00:47:47] - Big financial services functions that will be embedded in non-financial businesses

[00:51:07] - Deciding which functions and financial services models are most attractive

[00:57:01] - What a shift towards data and FinTech might unlock for the world writ large

[01:02:40] - How to improve payment profits by reducing credit rates

[01:04:12] - The threat that Buy-Now-Pay-Later companies pose to Visa and Mastercard

[01:12:17] - How the struggle between distribution and innovation continues to change

[01:15:04] - The kindest thing that anyone has ever done for him 


Copied from here:

[00:02:58] Patrick: Our guest today is Alex Rampell, a general partner at Andreessen Horowitz. Alex has a long history in FinTech, having co-founded six companies in his career, including Affirm and TrialPay. During our conversation, we cover Alex’s framework for positive selection in investing, why the best investments are often operating system or systems of record, and Alex’s views on the future of FinTech. For those that have listened to our Business Breakdown on Visa with Alex, you know the intellectual horsepower that he brings to every conversation and this conversation is no exception. Please enjoy my great conversation with Alex Rampell.

Lean into Positive Selection

[00:03:33] Patrick: So Alex, where the hell to begin this conversation? There’s going to be so many avenues we’re going to go down. Maybe I’ll start with a really simple concept I’ve seen you put out there. You call it the rules of investing. Rule number one is lean into positive selection and avoid adverse selection. Rule number two is just to see rule number one, so let’s zoom in on rule number one. What do you mean by this big overarching idea?

[00:03:52] Alex: Well, I think it’s particularly relevant for venture capital, because if you think about the world as a two by two matrix, there’s consensus and non-consensus and then there’s right and wrong. And if you’re in the wrong part, if you’re in the wrong column, you can never make money as an investor. By definition, this is the post-facto outcome. You were wrong. You can’t actually do well. The company that you invested in went bankrupt or lost to their eight competitors or whatever the case may be. But in public markets, you typically want to be non-consensus and right because obviously, you have to be right to actually make money. But if you’re consensus right, then that’s probably not that interesting. Whereas in venture capital, it turns out being consensus right is probably even better than being non-consensus right, because the challenge with being non-consensus right…

Again, you have to be right because if you’re wrong, not going to work. But the challenge with being non-consensus right is if you think, “Okay, this is a very, very interesting field. It’s going to require a hundred million dollars of capital for this business to actually get to the promised land and get enough customers and get enough revenue and get enough scale. I’m going to invest in their $10 million Series A, which means they’re going to eventually need to raise another $90 million in many successive rounds.” And this is a crazy, crazy idea that nobody except for me on planet earth believes in. Well, then who’s going to lead the Series B? This is the challenge. So it’s fine if you’re a non-consensus and then it trends towards consensus, but if the investment is non-consensus long enough, that’s actually great for public markets, typically. It’s not great for the private markets because you tend to need somebody else to believe and actually coalesce around what is your consensus.

And the other reason why public and private are just so different is there’s no concept of winning a deal in the public markets. You go to your interactive broker’s account or E-Trade or whatever and you enter in your order. You place a market order. You place a limit order. You place a stock order, whatever kind of order you place. You will get filled if you hit the ask. Whereas on the private side, a lot of times, there will be a company, it will only talk to five investors. You have to be in those five investors to be in the consideration set. If you’re the only one that’s actually bidding, again, that’s bad. That’s a sign of adverse selection potentially, because it’s the old Groucho Marx saying, if you wouldn’t want to join a club that would have you as a member, so you want to make sure that this is actually a hard to win deal, but then it typically is not about paying the highest price.

Because if it was paying the highest price, it wouldn’t be open to only five firms or ten firms. It’s just not as efficient of a market as the public markets. And in some cases, it’s not necessarily unfair. If you’re a private company and you want to fundraise in two days and you don’t want to leak your financials to potentially 10,000 investors that might tell all of your competitors, well of course, you want to actually tighten the circle and only talk to five firms. So the job is, again, to be one of those five firms. And it’s actually a great sign that those other four firms are highly, highly interested. It’s a great sign if you have to pay a much, much higher price than you would like, because it actually shows that there’s some element you might not be right. Right, wrong, not clear at that point, but there’s a heuristic, which is the fundraising acumen of the founder.

So I often say that the best founders can materialize labor and capital and the ones that are really, really good at materializing capital, again, it just shows that there’s a higher probability of the company getting sufficient funding to cross that chasm and get to profitability or get to sufficient scale. And that’s the high level of where I think about adverse selection. It’s the value investing trap has been very, very challenging in the public markets. If you followed value investing for the last 10, 15, probably 20 years, it’s even more problematic in the private markets because value investing means that something is wrong with the company. That’s why you’re getting a cheap price. And it’s very, very hard to turn a company around. It happens from time to time, but once you’ve lost your mojo and once you’ve lost your market cap, it’s very, very tough to turn that around because you have to pay people well. How do you do that if you have less money, if you have less market cap, again, private or public? So lean in to positive selection, avoid adverse selection. That is definitely a good rule to live by.

[00:07:53] Patrick: What do you think will happen in terms of capital formation and methods or technologies of capital formation in the private markets, especially early stage over the next 10 to 20 years? It seems like the prices, even though they seem really high, if it was just a pure open market and the most savvy storytellers could crowdfund their rounds or something, you can imagine prices that would get just ridiculous. What do you think might happen under those scenarios? Will we see anything like that? What are your thoughts on just more capital formation at the earliest stages?

[00:08:24] Alex: The answer is actually maybe. There have been plenty of times throughout recent history, where private markets are ahead of public markets. Now, why would private prices be ahead of public prices? Well, there are a couple of reasons for that. One is that you can’t shorten the private market, so there’s no selling of something that you believe to be overvalued. You just can’t really make that bed. But the other reason is by actually constricting the number of bidders, the bidders don’t really know that they’re overpaying, potentially. Whereas public markets, you’ll get a sense like, “Okay, here’s a company that wants to go do a secondary offering once public.” There’s already an auction that’s being conducted every nanosecond that gives a sense of what the price of the security should be and you don’t really have that in the private market. So there certainly is this capability for private markets to actually get ahead of public markets with a subset of the bidders that might exist on the public markets.

But I really love what this company Carta is doing because right now, you have this binary thing and normally, binary distinctions don’t make sense if you can have many, many different shades of gray. You’re either a public company or a private company. A lot of the value accretion has been in private markets, I think often to the unfairness of the retail investor or public market investors, because look at Amazon when they went public more than 20 years ago. I think it went public at a $550 million market cap or something very, very small like that.

[00:09:49] Patrick: Yeah, it was under a billion for sure.

[00:09:50] Alex: Yes, under a billion dollars was potentially a lot more money back then, just with inflation and whatnot. I’m not talking CPI inflation, but asset inflation .

[00:09:57] Patrick: Real inflation.

[00:09:58] Alex: Yeah exactly, real inflation of real assets, which has been pretty crazy of late. The vast majority of so many companies’ appreciation was in the public markets. Facebook went public at a hundred billion dollar market cap. And granted, Facebook is worth almost a trillion dollars now, so there’s been more net dollar accretion that way. But the idea that the public markets could actually participate in a lot of the upside and to your point, it should, over the long run, actually results in lower cost of capital for companies, which is a great thing for innovation and entrepreneurship. Because if you have this cartel of five investors that just do late stage deals, well, over time, they’re trying to arbitrage between public markets and private markets, which this was the way that a lot of late stage funds would make money for a very, very long time, is that they knew there was a liquidity premium and an illiquidity discount and that’s what they would trade on.

Another thing that I like to say a lot is that, and this is maybe a controversial statement so feel free to call me out on it, but there’s no such thing as an equity. There’s only a bond and a call option. And a lot of what private markets invest in, in particular, are out of the money call options. All these call options are not in the money or at the money. They’re very, very distinctively out of the money because how is a company that has two people in a garage and a PowerPoint presentation, how in the world is that worth a hundred million dollars? And it’s funny. My father, he’s an accountant. I often joke with him that when I talked to him about these things, it’s almost like Bud Fox talking to his father in that movie Wall Street, where Bud Fox is making money on these very fast and loose things. And my dad’s like, “When do these companies make a profit?

What about their profit? What about their profit?” I was like, “No, you have to understand these companies, it’s best at some point in time, they might be Facebook.” So you’re effectively betting on this is a very, very out of the money call option. If everything works perfectly, then it’s potentially the next Facebook. It’s potentially the next Apple and Apple looks more like a bond. Apple’s not an expensive company. Facebook’s not an expensive company, relative to everything else out there, because they’re no longer out of the money. But the vast majority of things early stage are out of the money. I know I’m jumping around a little bit here, but the reason why I like something like Carta, which is where I started this little monologue with, is they hopefully are enabling an exchange for people to invest in private markets so that you’re not… On day one, doesn’t make sense for your shares to be traded on NASDAQ and for you to put out quarterly reports when you have a product that doesn’t even exist yet. Nobody uses it. You don’t have a finance department and being Sarbanes-Oxley compliant doesn’t really make any sense.

Well, of course not. But a year before you go public, doesn’t make sense to only have trading in your stock, I don’t know, once every year and a half, haphazardly with only two firms that have a right of first refusal on your private securities, well, that probably doesn’t make sense either. What Carta has done is effectively, they are the cap table management software for almost every single private company out there. And now they’re enabling trading of those private securities with, again, not all of retail because you still have these accredited investor and qualified purchaser rules and a lot of other things that I think are rather anachronistic and should not exist, but that’s a different topic. But they’re blending the lines between private and public, which I think is the net great thing. Because actually when a company goes public, what should the reference price of its shares be? Well, it’s really hard to tell because a trade spawns once a year. It’s not even trading.

The way around and there’s somebody who’s buying it out of the money call option at that point. So it’s a much better idea to have a regular process and to have a non-binary process of that, which is you’re not private for nine years and then boom, suddenly you’re public and you have more shares traded in the first hour than your entire life as a company. No, transition to maybe you trade once a quarter. Maybe you trade once a month. You let more people on your cap table and hopefully the SCC loosens restraints around who can and cannot invest in private companies, not because I want people’s grandmothers taking on more risks, but it’s actually fundamentally unfair, from my perspective, that the value accretion goes to rich people that passed these accredited investor questionnaires.

Think in Terms of Bonds and Call Options, Not Equity

[00:14:02] Patrick: If we dig in a little bit more on the there is no equity, there’s only bonds and call options, how is that idea useful? It’s a fun thing to play with and it makes me think maybe of something like Amazon, where in the peak of Amazon’s reinvestment story, maybe over the last five years, that would be an example of a stock that I would have a really hard time parsing into bond because there is no bond. There’s no real profits to speak of relative to its revenue. And it’s reinvesting heavily, but it’s also very much a thing. AWS doesn’t feel like a call option. It’s working. It’s growing incredibly fast, even five years ago. So how did you come to this idea? Why do you think it’s useful for investors out there to think in these terms?

[00:14:40] Alex: Well, I mean, particularly for private companies, where it’s the only way to really think about it, you’re always taking a bet on the future of what if things go right? That’s the fun thing about private markets, which is what you can lose one times your money that’s capped. But if things really, really work, I mean, you could make 10,000, a hundred thousand, there’s unlimited upside. But how do you try to quantify that upside? And a lot of the ways that we try to do it, we know it’s false precision. Whereas if we say, “Okay, here’s a company that’s two days old. In year nine, we believe they will have $14.3 billion of EBITDA. Of course, that’s going to be wrong, but you try to look at what they’re building, not what is the total addressable market for that product or service as it stands today. It’s an out of the money thing, which is, if you think that the world is going in this particular direction, then let’s think about what that looks like at year five and year seven.

Wow, that’s a really, really big opportunity. So yeah, we’re going to say that this two people in a garage with a very, very fancy PowerPoint presentation, we’re going to call that worth $40 million or whatever the number is that the investors will put on it. But if that works and we own 20% of it and we get diluted over time, this could be a massive, massive market. We have to think about things in those terms because there are no profits. I mean, they’re just two people in a garage, never have profits definitionally so. I think for public markets, I mean, you could probably use the same framework, which is if every company is worth the present value of future profits, just almost objectively speaking, if you were going to go take a company private that’s currently public, that’s how you would have to value it to make your debt service and whatnot, so it turns out that the long tail of those profits is really what you care about. The discount that you apply to them is very, very high.

But if this thing is not working yet, but you think it’s going to work and be big in the future, then not really thinking about what year five looks like is probably not very smart. A lot of smart investors that I talked to, it’s almost like Seth Klarman has the famous Margin of Safety book, which is out of print. It’s like the original NFT. It’s very hard to get. But effectively, what you can do, again, with false precision, is you can try to apply a margin of safety to a wild, almost hyperbolic guess of if this thing really works, I’m going to invest in things that I think fundamentally can change the world. If they can fundamentally change the world in year five, I mean, any price that I pay today is effectively undervaluing this company relative to where it is in year five or year 10.

Otherwise, I wouldn’t be buying this out of the money call option. But I’m getting a margin of safety, if you will, of I’m only investing in things that I think can change the world that can be massive, massive, massive companies in year 5 or year 10. What’s a fair discount? What’s the margin of safety that I can get in year 10 if that thing happens? And again, it’s binary at that point. Either they succeed in that or they don’t. If they do, then you’re getting something very, very cheap. So if you invest in AWS or if you invest in Amazon when they first announce AWS and it’s like, “Wow, actually I think the entire world is going to be cloud.” Let me fast forward in 10 years at this company. Amazon gets 25% of the cloud market. Now, again, false precision. Are you going to get the exact size of the market? No, but you could almost apply a margin of safety there in that year 10.

That’s one way I think of coming up with a framework, which again, you have to have enough shots on goal to make this work. I mean, this is where the best venture firms, they’ll have concentrated bets, but you’re never going to put a hundred percent of your fund, I would argue, into one highly, highly speculative thing. It probably won’t work. That’s an out of the money call option. You’re probably hoping to say, “I’m going to invest in 10 things or 15 things or 20 things, or maybe 30 things that all have this chance of being completely revolutionary, world changing.” I like hanging out sometimes with other VCs because they tend to be very optimistic, in many cases overly so, around what can be world changing to what extent. Most of the time we’re wrong, like any other human, but if you get it right, that’s where, again, the out of the money call option that you’re getting is it turns out at the time to be very, very, fairly struck that in fact, unfairly struck in your favor because the upside is so big.

Searching for Operating Systems

[00:18:40] Patrick: Peter Thiel has this awesome definition of technology, which is really simple, which is just to do more with less. So this tool of leverage that creates for possibility and maybe therefore world-changing, to use your term. And one fun way I’ve heard you describe what you’re trying to do with your investing is that you’re hunting for operating systems, which are the ultimate form of technology leverage, if you will. Can you talk through this investment concept, what you mean by searching for operating systems as companies. And maybe we can go into as much detail as you’re able on this really interesting concept.

[00:19:12] Alex: My absolute favorite companies or businesses to invest in are ones that I think have an operating system like Mechanic. And that doesn’t mean that it’s Windows or Mac OS, but it has the same concept, which is if you ever go to a dentist or any modern dentist, I should say, almost every dentist in the country runs something called a DPM, a dental practice management software product. And that keeps track of all of the customers. It keeps track of the pictures of all of the customers’ teeth. And there are a number of companies that make them. Actually, one of the early ones was Henry Schein, which actually makes dental equipment, but turned out to get into the dental software space. But the retention rate of these products is basically a hundred percent. It’s a product that if you are the receptionist at a dental office or even the dentist himself or herself, you’re logging into this product every single day to check pictures of the teeth, to check when the next person’s appointment is, to check your outstanding billings, to go charge customers credit card. So it is the system of truth.

And when I say operating system, it actually means two things. It means system of truth. So it keeps track of everything at a company or even for a consumer, and I’ll talk about that a little bit later. It has very, very high utilization and usage. So it is this canonical, in consumer terms you would call it a DAU, a daily active use product or weekly active used product, because in the long run, what really matters the most, you could show evidence of a mode if you have a product that people use every single day and the margins that you’re able to extract from the product that you sell to these customers that use it every day, are maintained or even increased over time. So an operating system is basically something that runs the business, it is the system of truth, so it keeps track of what inventory you have, what your sales receipts are, how much you have in sales tax, all of these things. And the reason why that’s so valuable is because even though there is this kind of concept, the much valued concept of the App Store, you can start adding other things into that operating system.

So what does that mean? If you’re a dentist, you want to offer installment payments for the crown or cavity that that patient needs. It’s very easy, and you actually have free distribution of that new product if you are the operating system, versus what I would say, the very, very uphill battle of, “I am just a financing company that offers financing for cavities and crowns. Now I got to go find every dentist. I got to sign them up. The person that I signed up that works at the dental office might leave one month later, then I got to sign them up again. Then after I’ve signed them up, I have to hopefully count on the fact that they’re going to market this or push this in front of their patients, but they probably won’t do, so I have to re-market to them.” Again, versus the operating system where it’s the system of truth. There are operating systems for a lot of businesses like Toast, which went public recently. That’s an operating system for restaurants. They don’t just do payment processing. Like a lot of people think of it as like, “Oh, like I paid for my bill at the restaurant with Toast.” Well, Toast actually does payroll for the people that work at the restaurant.

They have tablets that go to the kitchen. So when the waiter or waitress goes and enters your hamburger order, it shows up immediately at the tablet at the kitchen. So it’s basically like a custom built piece of software that runs the business. And it will even keep track of how many hamburger patties are in the back kitchen as well. So these operating systems, they really retain customers extraordinarily well. And they are very adept. There’s a lot of what I would say out of the money call option value, if you will, of them being able to position other products and services to either the end customers, like the customer’s customers, or the customer itself. I actually wrote a blog post on this when I first joined this firm Andreessen Horowitz, which was my key learning, I called this the TiVo problem. This was at my company trial pay which I sold to Visa. So the TiVo problem, I call this, which is in 1998 TiVo and this other company called ReplayTV invented this amazing technology, at least amazing for people that were around it in 1998 like I was, that allowed you to pause live television. And TiVo was a very, very popular thing in the late nineties.

But today in 2021, it’s basically a patent troll rate. It was sold to another company which is effectively a patent troll that just sues other companies. I would never want to be in that position and I don’t have a lot of high regard for companies that do that. But the reason why that actually happened was TiVo did not control the distribution. They have this great product, but TiVo was not valuable if you just had a TV set and you lived in Antarctica. It only had value if you had Comcast, or if you had DirectTV. You need a TV to go. You need an actual television content to go into that TV, and then you would have live content to pause, hence TiVo. And I think the problem is that if you build an amazing, amazing innovation, and this is outside of the Clay Christiansen framework of disruptive versus sustaining innovations, it really is, “Do you control the distribution or not?”

So, Comcast has that pipe into your house. I think the problem is if you build TiVo, which is an amazing world changing thing, it’s not a sustaining innovation, it’s an amazing thing. But the problem is you have three outcomes that will eventually happen. Number one is Comcast says, “You know what? We should buy you. You’re an amazing company.” But if Comcast goes and buys TiVo, then what about Time Warner Cable and DirecTV? They’re going to say, “Hey, we’re not going to sell TiVo anymore. It’s owned by our competitor.” So you have this weird case in M & A where you can have not a control premium, which is a term often used where you’re paying more per share for the entire thing then you would for the marginal share. You’re going to have a control discount because TiVo is going to lose a huge chunk of their sales from the competitor. So that’s option one. Option two is that Comcast says, “Hey, you know what, let’s partner because we’re the ones that have the pipes into all the homes. We’re going to take 99 cents on the dollar. And you’re going to take 1 cent on the dollar.” And TiVo’s like, “Well, that’s not fair. I want a better deal than that.”

They’re like, “Yeah, well screw you. We’re just going to go with ReplayTV.” So you don’t really have that much leverage in a negotiation vis-a-vis the distributor. And then option number three is basically Comcast says, “That’s a nice little tool that you have there. We’re just going to go hire Accenture. I don’t know, some consulting firm or a bunch of engineers to go build a crappy version of the same thing.” And basically the problem is that one of those three options always happens to the TiVo, the metaphorical TiVo in this example, which is you build this amazing thing, it changes the world, you don’t control the distribution, unfortunately, and you either get copied, you get bought in an unfair price, or you get a partnership agreement which is really tilted out of your favor. So the lesson is, I mean, it sounds crazy to give this to an entrepreneur or a true innovator who’s like, “Don’t build TiVo, build Comcast.” Because if you build Comcast and you have a good product and engineering team, or you can actually create stuff, you have unlimited option value to go rollout TiVo, to charge more for TiVo, and so on and so forth.

Whereas if you’re TiVo, you’re kind of at the mercy of Comcast and you might get lucky, you probably won’t be. And 20 years later, you might get patent troll. And that’s kind of how I got to the operating system thesis to begin with, which is you want to look like Comcast. What does that mean? You want to be the pipes that actually control the backend of the business, because if you do that, and ideally even the front end. If you do that, you could be a body shop. Body shops should run on body shop software. Who’s going to build that software? Well, they’re going to have perpetual rights to offer, cross sell of whatever body shops need, whatever the customers of body shops need and so on and so forth. Or, you’ve got this whole other category of what I would call horizontal operating systems. QuickBooks is effectively an operating system. They do one thing for lots of types of businesses, which is the backend accounting. Or Square is a kind of operating system for lots and lots of businesses in a very horizontal way.

I use horizontal and vertical. As vertical is like focusing on one particular trant of business. It’s like Toast is a vertically focused operating system for restaurants, full stop. Square does that too, but it’s not as customized for restaurants, which is why Toast was able to steal a lot. But both of them effectively are operating systems. How do you know if it’s an operating system or not? I think this was a Supreme Court Justice Potter Stewart said, “How do you know if something is pornography?” And he said, “I’ll know it when I see it.” How do you know if something is an operating system? And I’ll say, “I’ll know it when I see it.” But really it’s like, “Is this thing the permanent system of record that stores all customer business interactions and is it used almost every single day?” And if the answer is, “yes,” it’s probably an operating system. If the answer is, “yes,” there’s almost this permanent up-sell capability where if you have, again, if you have a great management team, there are so many things that they can do with this.

Facebook is kind of an operating system for human interaction. That’s maybe a little bit of a stretch because there are plenty of ways of operating outside of Facebook. But what other products and features has Facebook added over the last 15 years? It’s really remarkable. So much of their business growth has been from that. Because again, they had very high retention, people use the thing almost every single day, and therefore there was a lot, like if you go add another feature, if you add a TiVo-like feature that’s really cool, you know that you’re going to get the distribution because you already have these daily interactions with customers.

[00:28:09] Patrick: Yeah, I absolutely love it. And I’ve got like six follow-up questions because I think we could kind of pick apart this as an investment strategy, almost the rest of the conversation. The first is around infrastructure and how you think about things like Amazon web services that are on bare metal or things like Twilio and Stripe, which are certainly underneath a lot and operating a lot of stuff on top of them, but typically are sort of abstracted away by developers so end users aren’t interfacing with them. So I guess another way of asking the question is, what do you think about infrastructure and how much does interface with an actual end customer matter in terms of this way of thinking?

[00:28:46] Alex: And I think it depends on what you’re trying to do. So I often call this the presentation layer conundrum, which is, I co-founded another company called Affirm that does point of sale lending. And if you look at payments online, there are very, very few companies that control the presentation layer. And presentation layer means they actually put something in front of the customer and they can actually change the graphics, they can change the text because they’re not relegated to the back end, or actually the front end. Paypal is one of the very few, I mean, Affirm does this as well. Afterpay does this as well. But most companies that do anything in payments, Visa’s worth $500 billion, so they must be doing something right. They are purely on the back end. They have no ability to actually change the interaction with the end customer. It limits what they can do. And that’s not bad because again, Visa’s worth half a trillion dollars, so they’re clearly doing a lot of things right.

But there are limitations on that because you basically have to assume that you’re going to be this dominant, middle-aware layer, do that as efficiently as possible, or become a protocol, or become like this dominant backend system of truth. But what I often talk about in this operating system thesis, which is, you tend not to have this optionality of product expansion, unless you own the customer’s interaction on the backend. Taking an example like Stripe, you use them as one of the ones that you quoted as like backend infrastructure. So Stripe has no interaction with the consumer. Generally has no interaction with the consumer. But Stripe, because they control the backend business, “I’m a business. I sell t-shirts online and there was a credit card form on my website. That’s how customers pay me. That’s processed with Stripe. The customers don’t know it’s Stripe.” Just like if I send them a text message, customers don’t know it’s Twilio, that’s all fine. “But I, as the business would rather get paid more quickly, or I would rather, I want to take out a loan secured by my future credit card receivables.” Well, because Stripe is the system of truth for me that knows what my chance of selling more t-shirts is next Tuesday, they still have product expansion capability with me on the backend.

And they can say, “I’m going to offer you, t-shirt company seller, a loan because I know it’s secured by these credit card receivables that were first in-line in, and it’s effectively a lien that I can exercise.” But it’s very, very hard if not impossible, because they don’t control the presentation layer, for Stripe to offer a service to the t-shirt buyer, if that makes sense? So a lot of this is really about expansion. There are very, very few companies that have the ability to kind of go what I would say, B to B to C, the business to business to consumer. Like Affirm actually was kind of conceived as an idea where we could get the end customers at Peloton because we were offering loans to them, and Peloton one of those to be branded as Affirm and not as Peloton loans. Because if you don’t pay your loan on time, like they don’t want to get mad at you, they want Affirm to get mad at you. They want to wash their hands of that responsibility.

I think in this framework, Twilio, if they play their cards right, and they clearly have, they have this ability to cross sell and upsell other products to the same customers, and they have the ability, and this is like the very, very rare thing, but they have the ability to scale with the scale of the business. Like the really cool thing about both AWS and Twilio is that yes, they could both face downward pricing pressure from competitors that offer the same services, but switching costs can be pretty high. So maybe Azure is cheaper this month, but how do I go switch all my servers over? And there are, of course, tools that do that, but it’s hard. But the really nice thing is that somebody like Twilio. Twilio goes and signs up Uber when Uber is one week old. I don’t know the exact story on this, but I assume it was something like this. And who would have thought that Uber would turn into this giant, giant business for Twilio? But of course it did. Or Amazon signs up this tiny company called Pinterest when Pinterest is one week old. Who would’ve thought that Pinterest would be such a giant AWS customer? But it became one.

It’s kind of nice when you’re able to retain the customer, grow with the customer, really scale with the customer’s success, which is what I mean by that, and then ideally you then control distribution to that customer. So if you happen to be privileged to have a great product team that builds new products and world-class products, you have instant distribution because you might not be the system of record, like AWS probably isn’t the system of record for Pinterest. And in fact, that doesn’t really make sense in that context for the context that they use the term, they’re not the operating system for Pinterest, but they’re the means of production if you will. They go build five new products. If those products are great, or even moderately okay, or just good enough, the chance of them getting that distribution is so much higher than a third-party that just shows up and says, “Hey, we’ve got a product that’s pretty good. Why don’t you go use it?” It’s too much work because you got to go struggle to get the distribution.

[00:33:33] Patrick: What have you learned about the sequencing to build one of these operating systems over time? Because when you take something to market, you’re not going to come with every feature. Toast came to market with something that’s narrower than what it currently offers its customers. So you have to sort of order things and choose things on the way to becoming the operating system and the system of record. What have you learned there? What are the most effective entrepreneurs do? And this may be different in horizontal versus vertical operating systems. What lessons for entrepreneurs that want to build an operating system in terms of sequencing?

[00:34:04] Alex: The sad truth is that even though you’re better off building an operating system than building a commoditized feature, you really need some kind of fundamental change in the world. It’s that kind of tailwinds that makes your operating system appealing. So what was one that happened recently? It was the shift to Cloud. “I’m running all of these things on my premises and my IT person was sick again, and my system went down and blah, blah, blah. I don’t want to deal with that again. I want to go switch to something that I don’t have to maintain and I’ll just pay for it monthly.” And this has been a decades long shift. I mean, there are still plenty of businesses that don’t run in the Cloud, but that was a massive tailwind for enabling, “I want to run Workday and not PeopleSoft,” or “I want to run NetSuite and not Microsoft Dynamics.” All of these shifts to the Cloud where one part of this enablement, because I don’t think that customers wake up and say, “Hey, I want an operating system that doesn’t really make sense.” They’ll say, “I want a feature.” Just like, I as a customer, by the way, would say, “I want TiVo. TiVo is awesome.” It’s very, very hard to lead with the whole ball, if you will, the whole operating system thing is just bad product marketing.

What’s good product marketing is, “Hey, do you want to loan?” And people say, they raise their hand and say, “Hey, I want a loan.” So to answer your question, what I’ve seen is the template that normally works is you lead with the very exciting feature that customers want, that’s different. You hopefully have your buttressed by this tailwind of a underlying platform shifts. So, “Oh, wow. Mobile has come out,” or “Wow, Cloud has come out.” And now you can do new things. Like Toast allows every waiter and waitress to carry around that tablet. I guess you could have done that 20 years ago with your PalmPilots or something. They wouldn’t have worked. It would have been terrible. So again, you’ve got this tailwind. You probably start off with one very, very narrow feature that you can do better. So for Toast, a lot of it was just credit card processing and payment processing. So you had to go upgrade your old credit card terminals to accept the chip and pin kind of thing, the EMV system. So you start with that, but then if you just do that, you’re probably running the risk of having a commodity service that’s going to get replaced by somebody who marginally under prices you, so you have to very, very quickly figure out, “How do I just retain this customer, serve as their operating system, add three or four other features as quickly as possible. And if I do, I’m probably in good shape. If not, I might kind of be screwed.”

This is the hard thing as an investor, by the way, which is, I’ll meet a company that says, “Okay, we’re going to do lending to small businesses by sending out postal mail to them.” I say, “Well, wow. Like somebody else is going to come along and send the same postal mail, but offer a 2% lower rate. And then you’re kind of screwed.” So how does this not become a race to the bottom? As far as entrepreneurs, in fact, all entrepreneurs that think about this problem, they’re like, “Well, here’s what we’re going to do. We’re actually going to do this first. We’re going to get them on our loan. Then we’re going to handle all of their bookkeeping for them because we built our own version of NetSuite, and we’re going to do this and we’re going to do that.” And it’s like, this is the Trojan horse, or this is what we often call in venture land, the wedge. This is the wedge that we’re using to pry ourselves into this particular business and aggressively expand.

And then the hard thing to really evaluate from that point is number one, it doesn’t matter what your plan is. Everybody has a plan. But what really matters is will the customer go for it? Is that an appealing value proposition for the customer? If yes, okay, great. And then number two is, can you actually execute on that quickly enough? And then number three, which is linked to number one and number two, is can you retain the customer long enough? Can you basically repel the competitive forces that lower rates long enough with enough stickiness or enough inertia, and inertia is often the thing that you need. The object that’s in motion is just kind of wants to stay in motion. The object that still wants to stay still. And both of those actually can work in your favor. But can you avoid these competitive forces long enough to allow your internal execution around product and engineering to go build these other products that you can then cross sell and hopefully become over time the system of record, versus I think what you were getting at, which is if you over-engineer and you spend five years in that garage with your co-founder and other members of your team building what you think customers might want, you have no idea.

It’s just very, very hard to figure out. So you’re better off with a wedge. And I think a lot of the best businesses at the early stage, they have an unfair path to getting what I would say, five customers. Because if you have an unfair path to getting one customer, because your uncle is somebody who owns something, well then you might over-engineer the product for that one customer and it has great applicability for that one customer, but no applicability thereafter. There’s a great Simpsons episode where Homer designs a car. It’s the stupidest car. It’s great for Homer, but it’s the worst car in the history of the world because it has no applicability for people that are not Homer Simpson. You run into that problem. If you’re able to do this with five customers that trust you, and this is really hard, by the way. If I were running a business, I’m running a dentist office, I’m a dentist, and my cousin says to me, “Hey, I’ve got this dental practice management software.” And I ask, “How much cash do you have left?” They’re like, “How long is your company in the ground?” I was like, “Well, I raised enough money. I had nine months in cash.” And it’s like, “What happens then?” “Well, we’ll probably have to shut down the cutlets.” I’m not going to run my dental practice on you. Are you crazy? It doesn’t make any sense.

If you could come up with some way of not tricking anybody, of course, but really selling this vision of what you’re going to accomplish. Come up with a wedge. Or even if you don’t come up with a wedge, get five people to fundamentally trust you, metaphorical five, it could be more, to trust you so that you actually are building a product around what the overall market wants, then you can really make one of these things work and you end up with a potentially phenomenal success. Toast is a great example of that. I don’t know all of the different steps that they took to get there, but it’s a really, really remarkable testament to what you can do when you become the operating system. And I think for companies like them, what other products and services do restaurants want? They’re probably 20 other ones, but now through the operating system, their customer acquisition costs on existing customers is zero. So they should be able to beat anybody who is the metaphorical TiVo in this example.

[00:40:12] Patrick: What about the rise of the creator class and the more individual, sole proprietor type businesses, online, digital only businesses? Is that in your view, the next tail end after cloud and mobile, that might mean we’re going to see a lot of interesting vertical software businesses that are operating systems?

[00:40:30] Alex: Absolutely. I don’t know if it’s the same technology or platform change. So I would say mobile and cloud were both things that new technology enabled this. Now, COVID tells me that I like staying home and I don’t like going to my office and I can do all these things on my own, but shoot, it turns out if I’m a lawyer and I already have 10 clients that love me and I know they’ll pay me, there’s a lot of stuff that I need to figure out on my own. And there’s an essay that we wrote on this called “The Rise of the Solopreneur”. So I want to be a lawyer on my own. Well, I already passed the bar. I already worked at Wilson Sonsini or something, so people know who I am. I’ve got 10 clients that love me that will follow me anywhere, but I need a website, I need to process payments, I need some kind of CRM tool. I need to be able to hire people. If I want an assistant, how do I do that? There’s a lot of work and it’s all these disparate, not connected systems.

So I love this idea of, I call it business in a box. So you want to go start a law firm. Well, obviously there was some things that a firm does, which are getting you clients. Maybe you could do that, but that’s not a software problem, that’s more of a marketplace problem or an advertising problem. But all of the other back-office things that can be done by software. And it’s very, very different for all of these fields. And what I think is interesting is that people decide to… You don’t really think of lawyers as being part of the creator economy, to your point, who are accountants or doctors. If I want to break out from the mothership, I just want to do the things that the firm did for me, but it’s going to involve 18 disparate systems and it’s a pain in the butt.

I think that’s definitely a trend. And it’s one that we’re very, very excited about. It’s the operating system, but very tightly targeted because the sole proprietor lawyer just has a completely different set of needs than the sole proprietor photographer or the sole proprietor yoga instructor. But there was a great round of software many, many years ago, and still today, mind-body sells to yoga studios, it performs a different function than people who used to work at a larger corporation or a larger entity, they just want to figure out how to branch off and break off on their own. And you just have to custom assemble the set of tools for that group of customers. The only challenge with this is, again, it goes back to the thing that I was talking about with the TiVo problem, customer acquisition is challenging.

But I think you have a much better chance of solving the customer acquisition problem if you have a very narrow set of, “This is Toast, it only does stuff for restaurants,” as opposed to quote-unquote, “We’re an operating system for you.” What does that even mean? It’s why I’ve been skeptical about a lot of tools that come out for what I would call creator economy or passion economy or sole proprietors without really individualizing the product. I’m very not skeptical about the rise of all three of those. It’s more of, what does it mean to be a solo worker? Well, it’s just like an Uber driver needs a completely different set of things for their tax filing and the number of miles that they drove and all of those things. That’s a product, but it’s very, very different than other verticals.

[00:43:33] Patrick: In addition to this awesome idea of the operating system, another thing obviously that you spend a lot of time thinking about is FinTech. And I’d love to turn the conversation there for a while. It’s where you do a lot of your investing, it’s where you founded businesses before. And maybe the right way to introduce our conversation on FinTech is with this funny joke you’ve got about the pig. Maybe you could give us the pig joke as an entry point into the world of FinTech.

[00:43:54] Alex: I love this one and I apologize for people that listen to this and they’ve heard me say it 10 times before. But basically the joke is there’re two pigs in a barn. One of them says to the other, he’s like, “This place is awesome. Everything is free, it’s heated, there’s free food, the water tastes great.” And the caption underneath says, “If you’re not the customer, you’re the product being sold,” which of course means that the pigs are being turned into bacon and they don’t even know it yet, but they’re living a life of luxury until they do. Basically, those were the two business models. Either you sell a product to a customer, and this is either a transactional business model or a subscription business model. So Peloton sells you a bike and they sell you a subscription, and you’re the customer. Or it’s the Facebook business model, which is you, the user of Facebook, are not the customer, you’re the product being sold. Hopefully the product that Facebook is offering is good, that’s why you show up. But the actual customer is the advertiser. And that’s where Facebook, where Google, draws in most of their revenue. So when we would meet a company, we’d say, “Well, which one are you? Are you an advertising company or are you a transaction company?” Because it was like bucket one, bucket two, there was no bucket three.

Now there is a bucket three and bucket three is effectively what I call embedded financial services. So now if you were to extend that joke, it turns out it’s like, “Oh no, the barn is free, we just have to use the checking account provided by the barn owner and hopefully use this debit card that has more than exempt interchange on it.” Et cetera, et cetera. I joined this firm in 2015 to spin up and run our FinTech practice. But now almost every company in some way, shape, or form is a FinTech company. Not because it is a pure play FinTech company, but because if you’re building the next Facebook, if you’re Mark Zuckerberg of 2021, you now see that there are three routes to revenue. You charge transaction fees or a subscription revenue to your customers, and you may sell advertising, and you might decide, “Hey, it’s very, very lucrative for me to offer financial products and services to my customers because they already trust me, they know who I am. And if I’m able to be the dominant checking account. If I’m their checking account,” which I know sound strange, you would think you get your checking account with Bank of America or First Republic or Chase or something like that. But if you have your checking account with somebody, they have so much control and ability, going back to the old refrain, to upselling products to sell you other things. So as an example, it might sound insane, but Uber and Lyft should offer checking accounts to all of their drivers for a few reasons.

One is it turns out both of those businesses are historically supply side constraints. So everybody wants to take an Uber from the airport at 5:00 PM, especially with all the stimulus checks and everything else that’s hitting, not as many people want to drive for Uber. They drive for Uber for two weeks then they quit. What would be very smart is we’re going to give them a checking account, and that has two benefits. One is when they’re running low on money, I can send them a message saying, “Hey, you’re low on money. Why don’t you drive for Uber today? We’ll pay twice as much.” It’s got this daily active use product. They don’t have to re-market to that customer because they already own the customer. And number two, the way that the whole, you’ve already listened to my visa thing because you interviewed for that, but for people that don’t know, the way that the credit card and debit card infrastructure works is that there’s typically something in the neighborhood of a 2% fee per card swipe, which is assessed to the merchant, which can be retained by what’s called the issuing bank or the issuer of the card.

So if Lyft gives every driver a card and a free checking account, and the free checking account is a lot more appealing than the one that Bank of America gives you that has minimum fees and all this crap. They give you this free thing, they own you as a customer, 2% of all the spending that you get, they get to keep, which is very compelling, and they get to win you back as a driver when you might be low on cash. And they’ve got that real retention at work. And again, you wouldn’t have thought of that as a use case 10 or 15 or 20 years ago. But now what we see is that even outside of what I would call the FinTech team, a huge number of enterprise software companies, and a huge number of consumer software companies, are trying to monetize with FinTech as a third leg of that stool.

[00:47:54] Patrick: Could you give maybe one example of that, that you’ve seen that you think is interesting? Because I’m most curious about what big functions, big financial service’s functions are most likely to start getting embedded in other non-financial businesses. So what’s an example that drives home the point?

[00:48:10] Alex: I think Toast is probably the clearest one, which is they offer payroll. If you’re a restaurant, why would you use ADP? Just use Toast. If you’re a restaurant and you need a loan because you know that you have working capital issues, Toast can actually go extend you alone. Toast of course, their primary source of revenue is interchange where they’re charging the restaurant something like 2.9% and the cost of actually doing that is probably a blended 1.85% or something like that. So they’re making that 1% plus spread on all the credit card processing. Toast is fundamentally not a credit card processor, they are a software company, which is very, very different than… There was a company like Toast… Well, they’re still around today, but it’s called Heartland Payment Systems. I forgot who they were part of now, it’s probably FIS or Pfizer or somebody, or WePay . But Heartland was basically, they were an independent sales organization that sold credit card terminals predominantly to restaurants. They didn’t have any software overlay. That’s why Heartland is not as relevant today as Toast is, and certainly not as big of a company, because Toast is the operating system that runs the restaurant.

That’s what they are first and foremost. If you ask a restaurant owner, that’s what they’re going to tell you, but they happen to make money via embedded financial services. That’s probably the clearest one. But the dental practice management software as well, there’s a company called Synchrony, they have a business called Care Credit, which is basically point of sale lending for elective medical procedures. And it’s predominantly for things like LASIK or dental or some kinds of dermatology and plastic surgery. And being embedded at the point of sale, or even before the point of sale, like the dentist, I went to the dentist, I sadly had a cavity. When I went to go check out and make arrangements to come back with my very painful filling, they said, “Hey, it’s going to be 400 and whatever it was dollars, but here’s a payment plan if you would like it,” because that’s embedded into the software that they’re using. And I’m sure they’re probably making more money on that. They’re probably keeping more of that than if it was just an upfront payment where they had to pay 2% to Visa MasterCard or the visa MasterCard system, if you will. I think there are a lot of examples of this.

I see it now with insurance, where a lot of companies are saying, “Hey, we want to bundle in.” When do you want to buy life insurance? Probably when you have a kid, when you buy a house, or when something very tragic happens to one of your friends, that’s when you might want to buy life insurance. So why not offer a life insurance cross sale if you’re a mortgage company? I wouldn’t be surprised if Zillow decides to say, “Hey, do you want life insurance now?” And maybe there’s a way of even lowering the interest rate on your mortgage. This will be a little bit complicated for Zillow to do. But that would be a great example where that’s a perfect non adverse selection point in time, so you don’t want to sell insurance to people that are planning on dying, of course, life insurance people. But you want to sell it when somebody is buying a house, how do you find that point in time? That’s an embedded financial service where if you can make the cross sell very, very easy and you have the distribution, again, very, very powerful.

[00:51:08] Patrick: How do you think about which financial service functions are the most attractive from a model standpoint? So the Toast is great because they get to enjoy a piece of all the transactions happening within their ecosystem. You’re basically riding commerce, you’re taxed on restaurant commerce at that point. Insurance is this more one-off example, it’s infrequent that people buy it, maybe it’s very valuable in terms of the individual order. What does that spectrum? How do you think about the different things that we will buy from these different companies that are financial services functions, and are those of the same as they’ve always been or are there new ones popping up?

[00:51:44] Alex: There’s a third element to that, which is can you underwrite better and more efficiently because you have more data? One of the things that actually is the most exciting to me around what could financial services accomplish, and I’ll go back to answering your question in a second, but I think this is pretty germane. Which is if you really wanted to revolutionize lending, what you would do is you would link it to payroll. Because I think in the US there’s something like $1.2 trillion, probably more now, of unsecured credit card debt. The average rate is 18%. Why is it 18%? Well, because you never know if these people are going to pay you back. There’s no enforcement mechanism outside of calling them a hundred times and saying they’re a bad person and reporting them to a credit bureau, which doesn’t always change people’s behavior. But where do people get their money from to begin with? Well, every two weeks, twice a month, depending on how they’re paid, they get paid from their employer, and then they are effectively a slow router and hopefully paying their bills on time and routing the paycheck that came in, the after-tax pay that came into their checking account to all of these different places. Wouldn’t it be really cool if you could actually get a loan from your employer?

Which sounds crazy, like, “I don’t want to take out a loan from the company that I work for,” but I would love it if I could… I need a better term for this than wage garnishment, because wage garnishment sounds very, very bad, but if I could say I want 5% of my paycheck sent automatically without my interceding at all to Capital One. And then capital one will say, “I will charge you 3% interest because I know you’re going to pay me every single week or every single two weeks or once a month or whatever, versus 18% interest.” There’s a powerful overlay as well, which is if you bundle these things more tightly, you’re often able to underwrite customers better. Because this is the problem. If I’m loaning money to a hundred people or a hundred businesses and I believe that half of them will not pay me back, I have to charge the entire group a hundred percent interest just to break even. Whereas if I can either underwrite them better or if I can link into just because of, not necessarily software, but just because of the stack that I control, where I see, I can link into, I can make sure that the money hits me first, I will be able to charge them dramatically less money.

So it’s not just, “Where are the limitations on this?” I think there are a lot of opportunities to unlock much fairer pricing. How many more people would buy trip cancellation insurance when you’re shopping at Expedia, which they offer, if it were really, really cheap based on your proclivity to either get sick or not? I just made that example up. But you can imagine where if you’re able to price things more intelligently either because you’re somehow in the flow of funds, you’re in the flow of data and it’s customized for that particular customer, I think you can actually broaden the scope of how this works. But going back to your original question, I think there is a very interesting thing to think about, which is, do customers want bundled experiences or unbundled experiences? So Robinhood is effectively, they’ve started bundling in other things beyond obviously core trading. If Chime offers trading, is that what their customers want or do Chime customers, Chime being the leading neobank in the US, do Chime customers want to have a Chinatown and a Robinhood account, or do they want to have a Chime account and a Robinhood account and maybe a lending account with blending club or upgrade, and do they want a life insurance account with some life insurance company and, and, and? Versus the bundled approach? I don’t know.

I think that’s a very, very good question. I think a lot of it, honestly, it goes back to inertia. If you can get them first, and this is the powerful thing, if by virtue of the fact that you’re bundling in the data, or you have access to the funds, you’re able to dramatically, dramatically price improve for the customer. That’s a very, very compelling reason. So for something like stock trading, it’s hard to be free or very low commissions, but for something like insurance, wow, that’s super interesting. Because if you have data that allows you to underwrite better, like you see this right now where Tesla has your driving score and Tesla’s starting to actually assure you. Which is pretty cool because Geico can save you 15%, but if you drive a Tesla and they know that you’re a safe driver and always follow the speed limit, which Geico does not do, it’s not because they don’t want to, they just don’t own the car. But theoretically Tesla should be able to under-price that. Or if Tesla was able to do something where it’s like, “Hey, link your…” Again, wage garnishment bad, but something that is a positive non pejorative form of that, “Pay us right away and your interest rate on your lease or your interest rate on your car purchase is going to be even lower.”

Which would be probably their captive auto finance company that would be doing that. But that would be very, very compelling as well. So I think there are a lot of areas for cross sell where, what excites me is the market opportunity could be much, much bigger when the distributors are, again, able to underwrite better based on data and potentially controlling the flow of funds, which is like the Toast example for giving the loan to the restaurant. The cheapest cost of capital for a restaurant would probably look something like a factoring arrangement. It’s called a merchant cash advance from the person that actually controls their funds. And normally you think of factoring as being more expensive, but in this case it could be a lot less expensive because the risk of loss can actually be very, very accurately measured. Versus what a bank has to rely on, which is, we have no idea, we have no bundled relationship with the customer.

[00:57:02] Patrick: I asked Dave Girouard at Upstart what his company might enable over 10 years, and he gave this cool answer, which is basically, anyone anywhere just has alone and terms presented to them at all times. And it’s based on all the data we have about them and they can just hit it or not. They can take the loan and terms are pre-calculated and you’re pre-approved and everything’s seamless and interesting. It’s fun to think about that applied to everything, not just lending, but all financial functions. If we think about finance existing because of basically the timing of cash flows, finance is just handling cashflow timing problems for other people, businesses, and individuals. If you think about it through that lens, what is most exciting to you about the future of FinTech? You get to see this from all angles, companies, big trends, et cetera, you’ve built companies in this space. What is the big, exciting future that this shift towards data and financial technology companies might unlock for the world?

[00:57:57] Alex: The timing of cash flows, it’s almost like this broken thing that doesn’t allow for customization, and therefore it requires people to go externally to go borrow money. Whereas I should be able to borrow money from… When this company Bill Me Later came out, I was like, “Well, that’s a really good idea. And actually it was inspiration in many ways for a firm many years later.” But I registered the domain name Pay Me Sooner, I still have that domain name, because I just think that there’s so many options where it’s like, “Wow, I have to borrow money because my customers are paying at 15. Those customers might even have too much cash, they’re very unhappy with the interest rate that they’re getting from their bank, or whoever’s providing that. Why can’t I offer them a small incentive in the form of a lower price for them to pay me sooner?” It goes in the pile of unused business ideas that I have. But actually there are a lot of these receivables, exchanges, and whatnot. But I think it’s one of many examples where, can you come up with… We started this conversation many minutes ago with DVRs, the digital video recorders. like replay TV and TiVo.

I wish you had that for money. By that I mean, why does a consumer get paid every… Not a consumer, but a worker, an employee, why do they get paid either semi-monthly or bi-weekly? That doesn’t make any sense. Well, it kind of does because the employer would rather pay every two weeks than pay every hour, but then that forces the employee to borrow money to maybe pay their rent on time. And that’s not fair, they should be able to customize those terms. And maybe the employer would be willing to customize those terms. Why doesn’t that happen? Well, there are many reasons. Sometimes it’s the employer’s cash flows, but in other cases, it’s competently complex. How do you manage all of this complexity? It was very, very difficult, I can imagine, in the pre Excel, pre software era to introduce, “Hey, you’re GE, you have 300,000 employees. Let everybody choose their own payday. That sounds like a bad idea. We’re not going to do that.” But why? That’s not that complicated, it’s just moving money around. And money is bits, it’s not atoms anymore, so just figure out that problem.

And now you have a massive unlock. You’ve actually exempted the need for exactly as you said, finances the, basically intermediating these differences in cash flows, and solving working capital issues. Now you don’t have one. You can actually solve this. And there are some that look like that. There are others where again, the reason why I use this grand experiment of lending is so inefficient today, because what is a credit score? It’s basically trying to figure out a business or consumer’s willingness and ability to repay. And generally, for businesses, if you actually have a Moody’s, Fitch, or S&P credit rating, it’s not about your willingness. You’re not going to just say, “Hey. I don’t really feel like paying my coupon. I’m just not going to do it.” But for a consumer, that happens a lot. It’s like, “Ah, I don’t really feel like paying Capital One. Screw them.” And you don’t know when that’s going to happen. They might get lazy. There are all sorts of reasons why a consumer might not pay on the willingness side. But ability is more mathematical. It’s not criminal. I shouldn’t say that, use a very broad term. But it’s really, really unfortunate that you’re not able to just make those a little bit more mathematical.

And that’s why I like, and I will come up with a better term than voluntary wage garnishment; but this idea that I, as a consumer, can choose where my money goes. And you basically take up the willingness score. If Capital One knew that everybody that they loan money to is going to pay them back, unless they are killed or fired, well, that takes out so many other different exigencies that they normally have to think about, which is, “Well, what if they just decide not to pay us because they decided not to pay us?” That’s the willingness part. Well, why not make this money movement more automated? And this is what smart contracts are on the crypto sense. “I know that the money will get dispersed to me, and therefore, I can offer a much, much, much lower rate.” And that’s the thing that’s kind of crazy in the world of finance today, which is, you have so many people that are seeking any kind of deal, because interest rates are zero, and that’s caused all sorts of craziness in the world that I’m sure you keep track of as well. And then, you have so many other people that are paying very, very high fees on all sorts of things.

And it’s like, “How do I reconcile the fact that those two worlds exist at the exact same time?” It doesn’t make any sense. And there are many areas that just have not truly been democratized; unless an instrument has truly been securitized, and it trades, and there’s a fair market for it, there are all sorts of these very, very esoteric things that are just one-off areas, where again, you can’t really figure out this willingness and ability score, if it’s anything around lending and credit. And then, lots of investors that would clearly lower the cost of capital for the other side, that would happily take a higher interest rate than zero that they’re getting with their bank. They have no access to that asset class, either. So I think there’s so many examples like that, which I’m just super excited about.

[01:02:41] Patrick: Is it too much of a stretch to think about this as, “There’s just too much interest paid in the world.” Or, “The profits of the financial services sector, if anything, the quality of life in the world, we want them to go down over time, because data is better and better. And there’s just smarter and smarter ways of doing this.” Is that the right way of thinking about this?

[01:03:00] Alex: Yes and no. You could imagine a scenario where the profits go up, because the rate goes down. And that’s actually the one that I’m more excited about. So you expand the pie dramatically, because there are lots of people that are locked out of credit right now. Or there are lots of people that decide not to use credit, because they don’t want to pay 18%. But they would happily take credit at 5%; but it’s all one size fits all, so they don’t do it. So, when you lower the price of something, you have your upward-facing supply curve, and your downward-sloping demand curve. And what you should see happen, is, “Wow. We’re lowering the cost of something. We’re making it much fairer.” And conceptually, you’re going to expand the pie dramatically.

But right now, it’s just too one size fits all, and in many cases, intransigent, in terms of just changing these things. So yes, I think it probably is good. That’s the hard part is again, it’s kind of prognostic in the future around, “How much bigger with some of these markets be if the pricing were really, really bare minimum, if you take out transaction costs?” And I’ve forgotten, there’s some famous economics study on this; but basically, if you take out transaction costs, you should see a lot more transaction volume. And then, there are all sorts of things that get built or get done, that just don’t happen today, because the surging cost of the transaction fees are just fundamentally too high.

[01:04:13] Patrick: You wrote a really interesting thread, and a good tie back to our original conversation on Visa, about the threat that “buy now pay later” companies may pose to companies like Visa. And this is just right at the beating heart of financial technology and finance, generally speaking. I’d love you to just summarize your thought process here, because Visa has been viewed by so many, and we explored it together, I encourage people to go listen, as one of history’s great business modes. And so, the concept that there may be innovation happening in this space that may pose a threat to longstanding, entrenched winners I’ll call them, is really interesting. Could you walk us through this concept?

[01:04:48] Alex: There have been many, many attempts to out Visa, or MasterCard and say, “I’m going to build a new payment mechanism.” But you really have to appeal to two different constituents. You have to make it very compelling to merchants. And it’s very easy to make anything compelling to merchants; just be like, “Hey. Do you want to pay 2% fees or 0% fees? How about zero?” And every merchant’s like, “Yay. I want to do that.” So it’s very easy to make it compelling to merchants. But then you have to make it compelling to consumers, and you say, “Hey. I’m going to give you no rewards, because I know you like getting rewards, but I’m going to give you none. Does that sound compelling?” And nobody raises their hand. But of course, the 2% fees the merchants are paying, most of that funds the rewards that consumers are driven by. So that’s one of many reasons why that mode is so high. So there have been many, many attempts to topple, or undermine, or figure out some kind of Trojan horse to disrupt this payment system; but nothing has really appealed organically to consumers or merchants.

There have been many things that appeal inorganically, by which I mean, Target has something called the Red Card; and because Target really hates paying Visa and MasterCard and that whole system 2%, they have decided to give consumers 5%, which is of course more than 2%, if they don’t use a Visa or MasterCard or American Express; rather, they use the Target Red Card. And it’s like, “Maybe that works. And they’re able to take away those 5% fees and make them zero.” But until they do, it’s an inorganic approach. So the thing that’s interesting about “buy now, pay later” is that it actually got organic traction with both consumers and merchants. So that’s reason number one. And why does it have organic attraction with both? There’s a famous clip, and you can look it up on YouTube, where Steve Balmer is pillaring the iPhone, saying, “Nobody’s going to buy a $700 or $800 phone.” And he was wrong, but he was right. He was right, because nobody was going to buy an $800 phone; but a lot of people would buy a phone if it was just $10 a month for the next 80 months on their AT&T bill, which is basically what happened.

So that’s why all these expensive smartphones took off. They were better products. But normally, better and more expensive doesn’t necessarily work; better and cheaper does. And that was kind of “buy now, pay later” for that space. It didn’t go on your credit card, where you’re going to get charged 18%. It was done by the principles involved; the AT&T and the Apple decided to go make this cheaper, and to go sell more units. Real clear benefit for the consumer, and again, real clear benefit for the merchant, because the merchant is going to sell more stuff. If they’re able to get out of the one size fits all financing regime that credit cards offer, which is credit cards have… They only see two different items that you’re buying. You’re either getting a cash advance, where you went to an ATM machine; and please don’t do this, because they’ll rip you off.

You use your credit card and not your debit card, they’re going to charge you like 35% on that thing. Or you are financing something that you bought at the store, in which case it’s, I don’t know, the 18%. They actually have no visibility, in terms of the item that you’re buying. So why was this good for merchants? Because merchants were able to customize and say, “Hey. We have too many of this item in our store. We want to charge 0% interest.” That was good for the merchant, because they moved more items if they’re charging 0%; this is why car companies have their Labor Day sales, with 0% APR. It’s good for the consumers, because consumers are going to buy something that otherwise they wouldn’t, because the cost was too prohibitively high. And yes, they had financing available, but it was very unfair financing. So it took off organically; little spiel on that as the background.

But what’s really, really interesting, I think, is that… So number one, it’s this new payment rail that has emerged, that it normally doesn’t touch the Visa or MasterCard rails at all. And then, this is the really exciting thing, which is, Visa and MasterCard, because they’re both so old, and this is not their fault at all, but because the system has five parties involved for a Visa or MasterCard transaction. You have the cardholder, me. You have the cardholders’ issuing bank, called Capital One. You have the card network, Visa. You have the acquiring bank, call it First Data or Bank of America. And then, you have the merchant, call it Chipotle. And those are the five parties involved in the transaction; and Visa sitting in the middle, has no idea what I bought at Chipotle. They assume it’s probably Mexican food because they’re smart, but they have no idea what actual skew or stock keeping unit I bought. And the same thing is true at Walmart. If I go buy something at Walmart, Walmart probably has millions of skews. If I spent $422 at Walmart, Visa has no idea what I bought. Capital One has no idea what I bought. And then, even First Data has no idea what I bought; because Walmart doesn’t want those three knowing what I bought, and the data infrastructure doesn’t even allow it.

That’s why, in many cases, a merchant name will get cut off. It will say you spent money at sq*, then the first 15 digits of the restaurant’s name. It’s just very, very antiquated technology. And it’s kind of hard to change, because it’s a standards-based protocol. How do you get everybody to just… All five of those parties, really not the consumer, but the other four to agree to use this more modern thing? So, one of the other cool things about the NPL, is that in order to underwrite effectively; Walmart doesn’t want to offer 0% financing on Purell wipes during the height of the pandemic, because they’re going to sell out of Purell wipes anyway. They want to offer 0% financing on… I don’t know, the old LG television, before the new one comes out, because they know they’re going to sell more if it’s 0% to the consumer, well, the merchant has to pay for that. That’s going to be a higher merchant discount rate. So they need to customize it based on the item being bought. The very interesting thing about the NPL. This is a long summary; my tweetstorm is a little bit shorter. But number one, it’s a parallel network that actually appeal to consumers and merchants; which sounds insignificant, but it’s pretty interesting, for a subset of transactions.

It doesn’t make sense to be NPL at $15 purchase. And then number two, is it has item-specific information. And you can do a lot of things with item-specific information, beyond just financing. You could say, “Okay. I want to apply a coupon. How do I do that? How does Visa apply a coupon? Or how does Capital One apply a coupon for an individual item that you bought at Walmart when they have no idea, if you actually bought that individual item?” The answer is they can’t. They can only offer basket-level discounts. So what the NPL is, is it’s the first parallel network that actually has gotten to scale, with consumers and merchants, that has item-specific information. And all it does is, very expensive purchases of $1,000 or more, that it divides up into different payments. It still has a good niche to be in, because yes, Visa launched theirs, and MasterCard launched theirs. But it’s never going to work, because they don’t know the item-specific information. Again, Walmart doesn’t want to discount Purell wipes with low-cost financing.

They want to discount, and pay for the discounting, I should say, the discounting being the suppressed, or I should say subsidized interest rate to the consumer. They want to do that for very expensive items that they would sell more of if they could offer better payment terms. It’s got item-specific information, and it has pretty good ubiquitous coverage and name recognition with consumers and merchants. And there just hasn’t been anything like that, at least in the offline world. PayPal arguably has done this, as well. PayPal is a parallel set of rails. A lot of PayPal transactions never touch credit cards or debit cards. But PayPal also fundamentally started off as a wrapper for credit cards and debit cards. I use PayPal all the time; but PayPal is effectively an abstraction layer that ends up charging my Capital One card. And that’s not the way that Affirm works.

Afterpay and Klarna are little bit more like PayPal does, because they tend to just charge your card. But what Afterpay does is pretty interesting; they effectively created what I would call a synthetic credit card from a debit card. So if you only have a debit card, can’t get a credit card. Now you kind of have a credit card, because your debit card is just getting charged four times over the next six weeks or so. But that’s kind of the big idea, which is, “What else can be done with this parallel network, given that merchants are not huge fans of the current oligopoly, the charges, and the very high fees?”

[01:12:18] Patrick: I think it’s the perfect excuse to sort of ask a question that puts a bow on the entire conversation, which is your idea about, you think about business writ large. It’s sort of the perennial struggle between distribution and innovation. I just love this simple concept, and I’d love you to summarize that concept in closing here. And also, maybe just riff on how you think that has changed across maybe your business and investing career, and how it might change in the future.

[01:12:43] Alex: This quote that I use all the time, is I said, “The battle between every startup and incumbent, comes down to whether the startup can get distribution before the incumbent gets innovation.” To clarify what that might mean, “Will Vanguard invent a robo-advisor, innovation, before, I don’t know, Wealthfront gets $5 trillion of assets under management?” That’s kind of the battle: which one of those two will win? And I think normally, in financial services, distribution is just so much more important than innovation, because it’s not to say that innovation is not important, but these are not atom-splitting things. These are not very, very hard things to build. You have to build a pretty good product, but the incumbent can fast-follow or slow-follow, as the case might be, and eventually build it. So it’s really informed my investment theory, which is, a lot of the companies that I invest in, I either like the infrastructure companies, where I’m not smart enough to figure out which one of the 400 neobanks is going to win, and get the most distribution. Or even if I want.

I’d just bought Google stock and Facebook stock or something, because that’s where they all advertise. But, if they’re all using Plaid, or if they’re all using Marketa, or they’re all using one of these infrastructure layers, the infrastructure layer actually becomes very interesting. So that’s one. Or, every now and then, you find some kind of very, very distinctive distribution wedge, where it’s like, “Wow. Like these guys built a good product; good for them. But they figured out a way to either make it organic, which almost never happens.” Where people just tell their friends. Yeah, you might tell your friends about some things, but you’re not going to say, “Wow. I love getting a mortgage at First Republic. It was such a great experience.” It’s not even relevant for most of your friends.

You’re not going to tell them. But if you come up with something, if you see something as an investor, that has just amazing, amazing, rapid adoption and distribution, that is very, very compelling. Rapid adoption of distribution is not like, “Wow. This company figured out how to buy Facebook ads really effectively.” Because somebody is going to figure out how to buy them even more effectively. It’s, they’ve figured out a different channel that cannot simply be exhausted as readily as what I would call these democratized channels; where it’s like, it’s great for the world that Google and Facebook allow anybody to be the highest bidder. But it’s not great for the long-term business value of anybody whose predominant source of customers and simply advertising on one of those two platforms. Because either the rules change, or somebody figures out how to go upside down on their economics; it just kind of imperils the source of oxygen, in the form of new customers, for the business.

[01:15:04] Patrick: Yeah. It’s a fascinating framework. I love thinking about things this way, simplifying some of the other frameworks that you’ve come up with. And as always, I’ve just learned so damn much talk to you about, not just FinTech, but just startups and innovation, more generally speaking. I have loved all the frameworks you’ve introduced today. I ask the same closing question of everybody that I have on this, the main show. What is the kindest thing that anyone’s ever done for you?

[01:15:27] Alex: Too many to list. I would definitely have to go with my parents, for creating me. So that’s a very kind thing. But there’s not just one case. But there are so many examples in Silicon Valley where I live, where you just go email somebody out of the blue; they’ll spend time with you, they’ll help you. Because it’s really not zero sum; this isn’t… As an entrepreneur, you’re not trying to figure out some arbitrage investment theory that somebody else is going to go steal from you, like in the way that hedge funds might not share ideas like this with each other. Maybe they do. I don’t know. But I mentioned, Max Levchin and I started the firm together. And that actually started when, I think I literally blind emailed him, saying, “Hey. I like you. I like your history of what you’ve done. Why don’t we meet to do this thing?” And many brainstorms later, we ended up becoming friends, and doing this.

But there are so many examples like that, where I was able to, I wouldn’t call it punch above my weight class, because I wasn’t punching above anything. I think it’s that line of Death of a Salesman, “I relied on the kindness of strangers” to really situate me here, because I moved here in 2004. I knew nobody. I moved here because my wife kind of talked me into it. I was from Florida. I didn’t want to move to California. And then, just a lot of kindness of strangers allowed me to meet some really incredible people, get some great ideas, and advanced my thinking. But I would have never gotten there, had I not just taken the initiative, but more importantly, had people not actually responded to my little easily deletable spam emails.

[01:16:48] Patrick: Wow, Alex, I’m so appreciative of everything you’ve taught me over the last couple of years, getting to know you. I really appreciate your time today. Thank you so much.

[01:16:55] Alex: Absolutely. It was great talking to you.